Millimeter-wave polarimetric radar system as an advanced vehicle control and warning sensor.

Abstract

By its invention at the turn of this century, the automobile has had a revolutionary impact on human society. Since the invention of the early automobiles, much effort has been devoted to providing a safer and more efficient transportation environment. The increasing importance of driving safety issues has initiated a great deal of research activities. The focus is both on minimizing injuries in such accidents as may occur and also in preventing the driving errors which may precipitate these accidents. Passive safety features such as seat belts and air bags were developed by the automotive industry, resulting in a significant reduction in fatal injuries due to accidents. Technology now becomes available to further reduce these accident statistics. Active safety features are designed to avoid accidents by detecting impending collision or unsafe driving conditions. Many of the new technologies are spin-offs from the aerospace and defense industries. However, the challenge lies not only in bringing these new technologies to market at affordable prices, but also in integrating them into the complex transportation environment to perform the expected functions. For example, the existing automotive radar sensors have been criticized for their intolerable false alarm rate. This disappointing performance is attributed to the lack of a systematic characterization of the traffic environment. Radar polarimetry is one way to characterize targets for identification. Besides solving the inherent problems in wave propagation and radar scattering, namely interference, multi-path, and signal to noise ratio, the polarization spectrum can be used to achieve target discrimination when targets fall within the same range gate. However, this application requires a thorough knowledge of the radar scattering behavior of different traffic targets, their associated interactions, and clutter. This thesis provides both theoretical and experimental approaches to the polarimetric characterization of traffic targets and clutter frequently encountered in the highway environment. Many scattering models are developed to accurately predict the backscatter behavior of road surfaces at near grazing incidence angles. The study examines the backscatter response of road surfaces of different varieties (asphalt and concrete) and surface roughness (smooth and rough), under various weather conditions (dry, wet, ice-covered, and snow-covered). Backscattering from debris and faults on road surfaces as well as from the roadside boundary is also covered in this investigation. The knowledge can be used to design safety features for automotive radar sensors; it also provides valuable information for adaptive cruise control applications. For the purpose of target categorization, a stochastic technique (the genetic algorithm) is developed to search for the optimum polarizations of transmit and receive antennas. This allows multiple target classification using a simple non-polarimetric radar. A prerequisite of this method is the a priori knowledge of the polarimetric response of the intended targets, which can be supplied by the theoretical models. A practical application of the genetic algorithm is the optimum design of an affordable non-polarimetric automotive radar sensor that can assess the physical condition of road surfaces.